Tefko Saracevic 1

criteria and methods in evaluation

of digital libraries: use & usability

Tefko Saracevic, Ph.D.

2

“Evaluating digital libraries is a bit like judging how successful is a marriage”

(Gary Marchionini, 2000)

(Gary with Chris Borgman – taken at the conferenceLibraries in the Digital Age (LIDA)held biannually in Zadar, Croatia)

Tefko Saracevic

3

ToC

• introductory musings• on the scene: several perspectives• what is needed for evaluation?• criteria for DL evaluation• methodologies for DL evaluation• the versus hypothesis• toward conclusions

Tefko Saracevic

4

evaluation: definition

Dictionary:assessment of value

the act of considering or examining something in order to judge its value, quality, importance, extent, or condition

in systems:assessment of performance in terms of effectiveness

and/or efficiency• effectiveness: how well did a system (or part thereof) do that

for which it was designed – related to objectives• efficiency: at what cost - $$$$, effort, time

Tefko Saracevic

5

evaluation …

in digital libraries:assessment of performance (effectiveness,

efficiency) on basis of given criteria • performance could be related to usability• criteria may be specified by users or

derived from professional practice, other sources or standards

• at issue:– what criteria to use?– what methods to employ?

Tefko Saracevic

6

digital libraries

• since emergence in early/mid 1990’s– many institutions & fields got involved– great many practical developments– many research efforts & programs globally– large & growing expenditures in practice– applications & use growing exponentially

• everything about digital libraries is explosive

• except evaluation– relatively small, even neglected area

Tefko Saracevic

7

literature reports on DL evaluation

• two distinct types:– meta or “about” literature

• suggest approaches, models, concepts; • discusses evaluation• useful for establishing a framework

– (e.g. Fuhr et al. 2007)

– object or “on” literature• actual evaluations, contains data

– (e.g. as included in Tenopir, 2003)

• but we are concentrating here on object literature only

Tefko Saracevic

8

boundaries of DL evaluation

• difficult to establish, apply – particularly as to process – e.g.

• crossing into IR: where does IR evaluation stop & DL evaluation start?

• or any technology evaluation? • or evaluation of web resources and portals?• is every usability study evaluation as well?

• brings up the perennial issues: – what is a digital library? what are all the

processes that fall under DL umbrella?

Tefko Saracevic

9

on the scene- as we discussed already

• several different communities involved in digital libraries, each with quite different– perspective, concepts, meanings in dealing

with DL– concentration, emphasis, approach, models– thus, different perspective in evaluation

• many disciplines, institutions involved– bringing different perspectives to evaluation

Tefko Saracevic

10

computer science perspectives:

emphasis in evaluation• concentrates on research & development

(R&D)

• technology centered– distributed & organized knowledge resources in

digital formats• how to collect, store, organize, diverse types of

information - texts, images, sounds, multimedia …

– new kind of distributed database services to manage unstructured multimedia resources

• and they want to evaluate those aspects

Tefko Saracevic

11

library & institutional perspective: emphasis in

evaluation• concentrates on institutions, service,

practice– logical extension of libraries

• content, collection, service centered– creation of digital collections– access to & use of collections– services provided

• guided by service mission• various environments, user communities• various degrees of integration or separation

• and they want to evaluate thatTefko Saracevic

12

organizational, subject perspective: emphasis in

evaluation

• variety of organizations involved– scientific & technical societies– various fields, academic units– projects - institutions, consortia– museums, historical societies– government agencies

• concentrate on collections & their uses in specific areas, subjects– new forms of publishing in their area

• services to communities or perceived needs

• and they want to evaluate thatTefko Saracevic

13

amount of evaluation in different communities

professional, subject organizations

library & institutional most

computer science least

Tefko Saracevic

14

what is needed to evaluate performance?

1. construct - system, process, part to be evaluated

2. objectives - reasons, desires for evaluation

3. criteria - standards, base for reflecting objectives

4. measure - units for recording & comparing criteria

5. measuring instruments - devices, tools that record a measure

6. methodology - way of doing evaluation• assembling, applying, analyzing

Tefko Saracevic

15

examples

Element Athletic event

Information retrieval (IR)

Construct 10 km race IR system, given IR method

Objective winner? effectiveness – how well did it perform?

Criteria speed - time relevance

Measure minutes, seconds

precision, recall

Instrument stopwatch people, judges

Method timing from start to finish

Text REtrieval Conference (TREC) laboratory

Tefko Saracevic

16

criteria in DL evaluation

• reflect performance of DL (or part) as related to selected objectives

– in studies: what parameters of performance were concentrated on?

• in DL: no basic or standardized criteria, no overall agreement

– many have been used– even for the same objectives

Tefko Saracevic

17

usability

• International Standards Organization - ISO 9241-11 (1998)

“Extent to which a user can achieve goals with effectiveness, efficiency and satisfaction in context of use”

• Jacob Nielsen (usability guru) definition: “Usability is a quality attribute that

assesses how easy user interfaces are to use. The word "usability" also refers to methods for improving ease-of-use during the design process.”

Tefko Saracevic

18

usability in DL

• widely used, but no uniform definition for DL

• general, meta criterion, covers a lot of ground

• umbrella for many specific criteria used in DL evaluations

Tefko Saracevic

19

six classes of criteria for DL evaluation derived from

literature• content

– how well are digital collections selected, developed; objects created, organized, represented, presented

• technology– how well do hardware & software support

library functions

• interface – what is available for users to interact &

how much is interaction supported or hindered

Tefko Saracevic

20

classes of criteria (cont.)

• process/service– what processes & assistance are provided; what

range of services is available; how well are they functioning; (carrying out tasks as: search, browse, navigate, find, evaluate or obtain a resource)

• user– what are the outcomes of DL use – changes in

human information behavior, cognitive state, decision-making, problem-solving; impact on accomplishing tasks; broader impact/benefit in research, professional work

• context– how well does a DL fit into, respond to, follow larger

context – institutional, economic, legal, social, cultural; effects on context

Tefko Saracevic

21

sample of criteria from literature

Content Technology Interfacecompleteness, sizecoverage, overlapquality, accuracyvalidity, authorityadequacy, diversityinformativeness freshnessaccessibility, availabilitycomplexity-organizational structure

transparency, clarityeffort to understand …

response timeprocessing timespeedcapacityloadaccessibilityeffectivenessefficiencycompatibilityqualityreliabilityrobustness…

attractivenessconsistencyrepresentation of concepts - labels communicativeness of messagesdisplay, attractivenessappropriatenessconsistencyease of useefforterror detection, personalization …

Tefko Saracevic

22

sample … (cont.)

Process/Service

User Context

learnability, effort/time, support, convenienceease of uselostness (confusion)completion (achievement of task)

interpretation difficultysureness in resultserror rateresponsivenessreliability,…

satisfaction, successrelevance, usefulness of resultsimpact, valuequality of experiencebarriers, irritabilitypreferenceslearning effectproductivityuse/reuse,…

institutional fit, usefulnessproductivity of & impact on community memberssustainabilityinteroperabilityrights management, copyright abidanceorganizational usability, …

Tefko Saracevic

23

criteria from Ying Zhang studyJASIST (2010) - from Rutgers PhD

dissertation

Content Technology InterfaceMore significant:

accessibilityaccuracyusefulnessfidelityintegrity

Less significant:conciseness

More significant:reliabilityease of useeffectivenessinteroperabilityefficiency

Less significant:flexibility

More significant:effectivenessease of useconsistencyeffort neededappropriateness

Less significant:personalization

Tefko Saracevic

24

criteria from Ying Zhang study …

Process/Service

User Context

More significant:reliabilityaccessibilityusefulnessresponsivenessintegrity

Less significant:courtesy

More significant:successsatisfactionuse/reuseproductivity

Less significant:behavior change

More significant:sustainabilitycollaborationrights managementmanagerial support

Less significant:extended social impact

Tefko Saracevic

25

methodologies

• digital libraries are complex entities– many methods appropriate– each has strengths, weaknesses

• range of methods used is wide– there is no “best” method– but, no agreement or standardization on

any methods

• makes generalizations difficult, even impossible

Tefko Saracevic

26

methodologies used

• surveys (most prevalent)

• interviews• observations• think aloud• focus groups• task performance

• log analysis• usage analysis• record analysis • experiments• economic analysis • case study• ethnographic

analysis

Tefko Saracevic

27

general results from all evaluation studies

• not synthesized here• hard to synthesize anyhow• generalizations are hard to come by• except one!

Tefko Saracevic

28

users and digital libraries

• a number of studies reported various versions of the same result:

users have many difficulties with DLs– usually do not fully understand them– they hold different conception of a DL

from operators or designers – they lack familiarity with the range of

capabilities, content and interactions– they often engage in blind alley

interactionsTefko Saracevic

29

a nice quote from an evaluation study

“It’s like being given a Rolls Royce and only knowing how to sound the horn”

quote from a surgeon in study of digital libraries in a clinical setting (Adams & Blanford, 2001)

Tefko Saracevic

30

analogy

• perceptions of users and perceptions of designers and operators of a DL are generally not very close

• users are from Venus and DLs are from Mars (or is it vice versa?) (nice NASA picture)

• leads to the versus hypothesis

Tefko Saracevic

31

is it:

• why VERSUS?– users and digital libraries see each other

differently

user AND digital libraryor

user VERSUS digital library

Tefko Saracevic

32

how close are they?user VERSUS digital library model

what user assumes aboutdigital library:how it works?

what to expect?

what digitallibrary assumes

about user:- behavior?- needs?

digital library model of user

user model of digital library

Tefko Saracevic

33

the versus hypothesis

in use, more often than not, digital library users and digital libraries are in an adversarial position

• hypothesis does not apportion blame– does not say that DL are poorly designed – or that users are poorly prepared

• adversarial relation may be a natural order of things

Tefko Saracevic

34

toward conclusions:evaluation of digital libraries

• impossible? not really• hard? very• could not generalize yet• no theories• no general models embraced yet,

although quite a few proposed• in comparison to total works on DL,

only a fraction devoted to evaluation

Tefko Saracevic

35

why? – some speculations

• complexity: DLs are highly complex – more than technological systems alone– evaluation of complex systems is very hard– just learning how to do this job – experimenting with doing it in many different ways

• premature: it may be too early in the evolution of DL for evaluation on a more organized scale

Tefko Saracevic

36

why? (cont.)

• interest: there is no interest in evaluation – R&D interested in doing, building, implementing,

breaking new paths, operating … – evaluation of little or no interest, plus there is no

time to do it, no payoff• funding: inadequate or no funds for

evaluation– evaluation time consuming, expensive requires

commitment – grants have minimal or no funds for evaluation– granting agencies not allocating programs for

evaluation– no funds = no evaluation.

Tefko Saracevic

37

why? (cont.)

• culture: evaluation not a part of research and operations of DL– below the cultural radar; a stepchild– communities with very different

cultures involved • language, frames of reference, priorities,

understandings differ• communication is hard, at times

impossible

– evaluation means very different things to different constituencies

Tefko Saracevic

38

why – the end

• cynical: who wants to know or demonstrate actual performance? – emperor clothes around? – evaluation may be subconsciously or

consciously suppressed– dangerous?

Tefko Saracevic

Tefko Saracevic 39

ultimate evaluation

• the ultimate evaluation of digital libraries:– assessing transformation in their

context, environment – how did DL affect them?

– determining possible enhancing changes in institutions, learning, scholarly publishing, disciplines, small worlds …

– and ultimately determining effects in society due to digital libraries

Tefko Saracevic 40

final conclusion finally

• evaluation of digital libraries still in formative years

• not funded much, if at all• but necessary for understanding

how to – build better digital libraries &

services & – enhance their role

41

evaluation perspective –Rockwell

Tefko Saracevic

42

still another one …

Tefko Saracevic

Tefko Saracevic 43